Selection of Bioactive Ionic Compounds

The group of compounds selected to demonstrate the role of interaction mechanisms is based on hierarchy and representation. On the one hand, we selected three representative levels of complexity regarding the number of biofunctional groups in either the anionic or cationic structural subsection (Scheme 1). On the other hand, it was important to select tunable motifs in which chain length, chirality, aromatic vs. aliphatic moieties, etc. can be systematically varied with a clear knowledge of the role of these structural elements in terms of bioactivity.^30-32 The bioactive compounds are classified into three groups based on their structural characteristics.

The synthesis of alkyl [(1R,2S,5R)-(−)-menthoxymethyl]dimethylammonium chlorides (C_xAmOMCl) is shown in Figure 1a. Three chiral quaternary ammonium salts with chloride as the anion and with even-numbered alkyl chain lengths in the cation part—namely ethyl (C₂AmOMCl), hexyl (C₆AmOMCl), and dodecyl (C₁₂AmOMCl)—were prepared via a two-step approach under mild synthetic conditions (Figure 1b). (1R,2S,5R)-(−)-Menthol was selected as the primary building block for the cation because it is a naturally occurring compound with a wide range of bio-applications in fields related to medical^{40, 41} and biotechnology.⁴² A detailed description of the quaternization reaction and characterization of the resulting C_xAmOMCl is provided in the Supporting Information. The physicochemical properties of the obtained quaternary ammonium salts revealed these salts²⁹ to be ILs.⁴³ Nuclear magnetic resonance (NMR) and mass spectra of the C_xAmOMCl products (Figure S1–S19) confirmed the successful synthesis of all chiral ILs (CILs).

The homologous series of C_xAmOMCl presented here exhibits a broad range of biological activity. These compounds are effective as antimicrobial agents against bacterial rods, cocci, bacilli, and fungi.^{29, 30}

We also observe an enhancement in the biocidal effectiveness of C_xAmOMCl against pathogenic bacteria and fungi in the presence of the (−) enantiomer of menthol in the structure; such an enhancement is not noted in their racemic counterparts.³⁰ Additionally, it is worth mentioning that biotests reveal a strong correlation between the chain length and biological activity. C_xAmOMCl salts with long alkyl chains (ranging from 9 to 12) engage in much higher biological activity than salts with shorter alkyl chains (ranging from 1 to 8). Notably, the biocompound selected as a prominent model system, dodecyl[(1R,2S,5R)-(−)-menthoxymethyl]dimethylammonium chloride (C₁₂AmOMCl), demonstrates significantly higher antimicrobial activity compared to the commonly used benzalkonium chloride (BAC). For most of the Gram-positive and Gram-negative bacterial strains and yeast-like fungi tested, the minimum inhibitory concentration (MIC) and minimum bactericidal/fungicidal concentration values for C₁₂AmOMCl are significantly lower than those for BAC.³⁰ For example, the MIC against M. luteus for C₁₂AmOMCl is 0.5 μM, compared to 1.4 μM for BAC, and for E. faecium, the MIC is 0.5 μM for C₁₂AmOMCl, compared to 5.6 μM for BAC. This result is particularly promising because long-term use of BAC can lead to BAC resistance in various microbial isolates. These findings suggest that C₁₂AmOMCl could be highly effective for antisepsis and disinfection purposes.

Furthermore, our findings reveal that CILs with (1R,2S,5R)-(−)-menthol featuring an aliphatic ammonium core exhibit the broadest spectrum of the most potent activity against microorganisms compared to other CILs with (1R,2S,5R)-(−)-menthol containing imidazolium or pyridinium cores. In our previous work, we found that C₁₂AmOMCl has antifungal and antiadhesive properties against C. albicans.⁴⁴ Our results reveal that this CIL is able to permeate the membranes of C. albicans and degrade their cell walls. For a compound to be considered a viable disinfectant, it must not cause hemolysis. Therefore, we tested the hemolytic activity of quaternary ammonium chloride with a dodecyl alkyl chain at its active concentrations for killing fungal cells, detaching adherent C. albicans cells from surfaces, inhibiting filamentation, permeabilizing membranes, and disrupting cell walls.⁴⁴ Our results showed that C₁₂AmOMCl has very low hemolytic activity, ranging from 3–12 % at concentrations of 10–100 μM, with 1 % sodium dodecyl sulfate serving as a positive control for 100 % hemolysis. The C₁₂AmOMCl we investigated has potential as a disinfectant because it not only has antifungal and anti-adhesive activity, but also does not induce hemolysis.

Adsorption Characteristics of Ammonium-Based Cations in Nanoporous Carbons

The calculated dimensions of the cations are summarized in Table S1.⁴⁵ The pore size distribution of the chosen model carbon NORIT ROX 0.8 (termed ROX) is broad and ranges from 1–10 nm (Figure S20). The cation length of the selected bioactive ILs ranges from 0.86 nm (choline cations) to 2.48 nm (C₁₂AmOM cations),⁴⁵ which makes it possible to study ionic structures and adsorption behavior in ROX carbon electrodes. We investigated varying alkyl chain lengths to examine the interaction between the hydrophilic group of the bioactive ions and the surface of the carbon electrodes.

We evaluated the interaction between the selected bioactive ions and nanoporous carbon by measuring the zeta potential of carbon in suspension solutions; the carbon materials were treated with 0.1 M bioactive ion aqueous solutions (see Figure S21 and the experimental section for details of the zeta potential measurement). As shown in Figure 2a, the zeta potential for carbon with pure water treatment is roughly −43.6 mV, which reveals the negatively charged surface of ROX. The zeta potentials for acetylcholine chloride (AChCl)- and ChCl-treated ROX carbon are −47.6 and −21.4 mV, suggesting a weak interaction of acetylcholine and choline cations with ROX. The zeta potentials of other bioactive ion-treated ROX carbons exhibited an obvious increase and changed from −9.5 mV (C₂AmOMCl), 3.1 mV (C₆AmOMCl) to 41.5 mV (C₁₂AmOMCl) as the alkyl chain increases from ethyl to dodecyl. This finding indicated increasing physisorption with increasing alkyl chain length.

In Figure 2b we compared the adsorption-related properties of bioactive ions in porous carbon electrodes. Relying on an in situ-Raman setup (Figure S22), we finely calibrated the bioactive ion concentration and the Raman intensity (Figure S23 and S24). For ChCl and AChCl (ChCl- and AChCl-treated ROX carbons with negative zeta potentials of −21.4 and −47.6 mV, which indicates a weak interaction between the cations and nanoporous carbons), the measured concentrations exhibited a slight decrease after 13 h (5-h of physisorption, 4-h of electroadsorption, and 4-h of electrodesorption; Figure 2b); the electrosorption-induced concentration changes in a 0.01 M ChCl electrolyte have been confirmed via ex situ-liquid chromatography-mass spectrometry (LC-MS).¹¹ We noted a clear trend for stronger adsorption kinetics with increasing alkyl chain length according to our in situ-Raman results (the fluctuations in concentration are mainly due to the limited signal-to-noise ratio of Raman spectroscopy under in situ-testing conditions). For C₁₂AmOMCl, there were only weak C₁₂AmOM signals after 5 h (physisorption) according to the Raman spectra, and no signal was observed after 13 h (Figure S25). The adsorption kinetics increased significantly from C₂AmOMCl to C₆AmOMCl, and the corresponding zeta potential values changed from −9.5 mV (C₂AmOMCl) to 3.1 mV (C₆AmOMCl) (Figure 2a–b). C₁₂AmOMCl exhibited similar adsorption behavior in the ROX electrode as C₆AmOMCl. The zeta potential data and the in situ-Raman spectra jointly indicated that the hydrophobic interaction depended on alkyl chain rather than on the presence of the monoterpene moiety; furthermore, the physisorption intensified with the increasing alkyl chain length. Notably, C_xAmOMCl (where x=2, 6, and 12) that strongly physically adsorbed in nanoporous carbons could not be desorbed by electric polarization (Figure 2b).

We also treated ROX powders with 0.1 M ChCl and C₁₂AmOMCl solutions (washed with pure water three times) and termed the final products ROX-ChCl and ROX-C₁₂AmOMCl. These products were characterized via N₂ physisorption at 77 K to evaluate the porosities of ROX after physisorption with ChCl and C₁₂AmOMCl (Figure 2c). We noted almost no decrease in SSA of ROX-ChCl (1162 m² g⁻¹) compared to the initial ROX (1182 m² g⁻¹). However, we observed a strong decrease in SSA for ROX-C₁₂AmOMCl (198 m² g⁻¹), which indicated a strong interaction between C₁₂AmOM cations and ROX carbon powders. Given the size difference between ChCl and C₁₂AmOMCl, the porosity difference may have also resulted from the large cation size of C₁₂AmOMCl. Therefore, we used the independent LC-MS experiments to confirm that the amount of physically adsorbed C₁₂AmOMCl (up to 200 μmol/g after 5 h and 562 μmol/g after 48 h) was hundreds of times higher than that of ChCl (less than 5 μmol/g after 48 h), as shown in Figure 2d.

We employed ex situ-¹H MAS NMR spectroscopy to explore the physisorption of C₁₂AmOMCl on ROX carbon powders with a broad pore size distribution at the molecular level. The well-known ring current effect in graphitic carbon materials decreases the chemical shifts of adsorbed molecules.^46-50 By studying those so-called nuclear-independent chemical shifts (NICS), we were able to discriminate between molecules within the pores of the carbon materials and those in the surrounding bulk electrolyte. ¹H MAS NMR spectra of ROX loaded with defined amounts of 0.1 M C₁₂AmOMCl ²H₂O solutions relative to the available pore volumes (V_pore) of ROX are shown in Figure 2e and 2f. For the samples shown in Figure 2e, we conducted measurements immediately after we had loaded the carbon material and did not allow for any additional equilibration time except for the sample transfer time necessary to move the sample into the spectrometer (ca. 10 minutes). The alkyl groups of C₁₂AmOMCl are well resolved in the bulk phase and occurred between 1 and 5 ppm. They exhibited negative chemical shifts and two poorly resolved maxima at −3.8 and −6.7 ppm for the three different loading volumes, i.e., the signals exhibited severe line broadening in addition to a negative shift. The negative chemical shift resulted from the ring current effect noted above; it revealed that all of the C₁₂AmOMCl molecules were adsorbed within the pores immediately after loading. The pronounced line broadening is caused by the immobilization of the adsorbed molecules within the pores; this broadening might be a sign of different states of the adsorbed molecules, which could have resulted in a distribution of chemical shift. After 72 h of equilibration (Figure 2f), the lines of the adsorbed C₁₂AmOMCl molecules at negative chemical shifts became somewhat narrower and were then centered in one primary peak. This result attested to a slow redistribution of cations within the pores of the ROX carbon. A larger amount of C₁₂AmOMCl electrolyte resulted in more negative chemical shifts (i.e., −3.7 ppm for 0.5 V_pore, −4.1 ppm for 1.0 V_pore, and −5.5 ppm for 2.5 V_pore). In summary, ¹H MAS NMR spectroscopy revealed a rapid uptake of C₁₂AmOMCl by the ROX carbon material. However, the adsorbed molecules rearranged themselves slowly within the material over a longer period of time.

The mechanism of C_xAmOMCl interaction with ROX carbon was further evaluated by comparing C_2-OHAmOMCl and C₂AmOMCl. The presence of the OH group in C_2-OHAmOMCl weakened the adsorption of molecules in porous carbons with hydrophobic surfaces, which also confirmed that the interacting group of the bioactive cations for the exchange of C₂AmOMCl to C₁₂AmOMCl was the alkyl chain rather than the monoterpene moiety (Figure S26–36 and Section 2 in the Supporting Information). Additionally, all of these CILs we investigated were electrochemically stable under electric polarization (Figure S37–S39).

Investigation on the Synergistic Adsorption Effect of Ions in Nanoporous Carbons

We have shown that, based on the interaction behavior of bioactive ammonium-based cations with nanoporous carbons, that a longer alkyl chain led to stronger irreversible physisorption. We selected six typical organic compounds, C₁₂AmCl, sodium 1-decanesulfonate (NaC₁₀SO₃), NaC₈OO, NaSal, NaIbu, and dopamine hydrochloride (DopaCl), to further investigate the ion physisorption behavior in ROX carbon (Figure 3a) via Raman spectroscopy (see the calibration curves in Figure S40). C₁₂AmCl and C₁₂AmOMCl exhibited similar interaction behaviors with ROX carbon, which suggested that the alkyl chain length played a key role in the strong physisorption of cations with ROX carbon (Figure 3b).

In the context of other bioactive ionic compounds that do not belong to quaternary ammonium salts, we observed a variety of interaction behaviors of anions with long alkyl chains in NaC₁₀SO₃ and NaC₈OO; however, we observed only a slight decrease in concentration for NaC₁₀SO₃ and NaC₈OO compared to C₁₂AmOMCl after 5 h (Figure 3b). For these compounds, a long alkyl chain was apparently not a sufficient and necessary condition for strong and irreversible physisorption of C_xAmOMCl; physisorption is caused by the synergistic effect of hydrophobic and electrostatic attraction (Figure 3c). ROX carbon is a form of activated carbon with negative surface functional groups (such as -O⁻ and -COO⁻).⁵¹ It has a negative zeta potential (−43.6 mV) (Figure 2a and 3c). C₁₂AmOM cations will accordingly adsorb on the surface of ROX via electrostatic attraction, and then the hydrophobic alkyl chain will be attracted to the carbon surface via hydrophobic interactions. That situation leads to strong and irreversible physisorption between ROX carbon and C₁₂AmOM cations. However, for anions with long alkyl chains, such as C₁₀SO₃^- and C₈OO^- anions, the electrostatic repulsive force is responsible for suppressing the adsorption on the negatively charged ROX carbon surface, which results in relatively weak physisorption compared to that observed for C₁₂AmOM cations.

We next investigated three commercially important and common bioactive ionic compounds—NaSal, NaIbu, and DopaCl—to examine the theory of the synergistic adsorption effect via in situ-Raman experiments (Figure 3b). The salicylate and ibuprofen anions exhibited smaller concentration changes than C₁₂AmOM cations after 5 h of physisorption. Dopamine cations with two OH groups exhibited slightly stronger interactions with ROX carbon than salicylate and ibuprofen anions. Negative zeta potentials are typical for ROX carbon and other common nanoporous carbons, and the hypothesis was also verified for YP-50F and AC-CS carbons (Figure S41 and Section 2 in the Supporting Information).

Irreversible-Physisorption Induced Confined Cations and Relative Ionic Memresistance

Due to the irreversible physisorption, C₁₂AmOM cations were strongly adsorbed and confined by ROX carbon; chloride anions remained mobile in the solution. During electric polarization, C₁₂AmOM cations were fixed on the surface of the polarized carbon electrodes; the chloride anions were adsorbed and desorbed to compensate the charges (Figure 4a). Confining the C₁₂AmOM cations resulted in anion enrichment on the positive electrode and anion depletion on the negative electrode during electric polarization, which led to memristive behavior. For the weak interaction mode with ChCl, both choline cations and chloride anions contribute to balancing the opposite charges (Figure 4a). Benefiting from this phenomenon, we developed the all-carbon ionic memristor shown in Figure 4b. It is composed of an outer main capacitor (M-Cap, with two large ROX carbon electrodes) for ion electrosorption and an inner detective capacitor (D-Cap, with two small ROX carbon electrodes) for detecting the ions’ dynamics and relative changes in resistance. We explored different electrolyte systems to study the resistance changes of the D-Cap while deliberately controlling the M-Cap voltage (Figure 4c–f). We applied ten 1 V-pulses followed by ten 0 V-pulses with the same pulse time (t_p=20 s) to the M-Cap in a 0.1 M C₁₂AmOMCl electrolyte (Figure 4c). We evaluated the resistances of the D-Cap using electrochemical impedance spectroscopy (EIS) during the rest time (t_r=20 s, no bias voltage was applied to the M-Cap; only the open circuit voltage was recorded).

We noted a 15 % and 85 % increase in resistance of the D-Cap (compared to its initial resistance) after 1 and 10 times of 1 V-pulses. We also compared the resistance changes for the D-Cap in C₁₂AmOMCl and ChCl electrolytes (Figure 4d). Due to variations in interaction behaviors of C₁₂AmOMCl and ChCl with ROX carbon, we observed pronounced resistance changes in the D-Cap in 0.1 M C₁₂AmOMCl and ChCl electrolytes. While we observed an 85 % increase in resistance of the D-Cap after the first ten 1 V-pulses in a 0.1 M C₁₂AmOMCl electrolyte, only a 6 % increase was noted in a 0.1 M ChCl electrolyte. The actual values of resistance changes and impedance spectra are shown in Figure 4e and S42. That difference is due to variations in the numbers of confined C₁₂AmOM cations and weakly adsorbed choline cations; the confinement of C₁₂AmOM cations—additionally led to anion enrichment and anion depletion during electric polarization (Figure 4a). In addition, we observed a 9 % increase in resistance for the first ten 1 V-pulses in NaC₁₀SO₃, which has a weak interaction with ROX carbon even with a long chain in the anionic moiety (Figure S43). We further evaluated the switching energy of these 4-terminal devices in C₁₂AmOMCl and ChCl electrolytes by integrating the area of I-t curves of M-Cap. The average values of switching energy for 1 time of 1 V-pulse are 504.8 and 97.1 μJ/mm², for 0.1 M ChCl and C₁₂AmOMCl electrolytes, respectively (Figure 4f). By decreasing the mass loading of M-Cap (from 60 to 10 mg) and concentration of C₁₂AmOMCl electrolyte (from 0.1 to 0.01 M), the switching energy decreased from 97.1 to 1.5 μJ/mm² for 1 time of 1 V-pulse (Figure S44). The memory time of this 4-terminal ionic memristor was investigated and the resistance state of D-Cap after 1, 10, and 100 times of 1 V-pulses to the M-Cap was evaluated in Figure 4g and S45. After one 1 V-pulse, there was an 18 % increase in resistance of the D-Cap, and the resistance gradually decreased to 7 % after 240 s. After ten subsequent 1 V-pulses, we noted a 75 % increase in resistance of the D-Cap, and a similar decrease in resistance (to 45 %) was observed after 400 s. When increasing pulse times to 100, there was a 293 % increase in resistance of D-Cap, which decreased to 216 % of the initial resistance after 4000 s. These results indicate a nonvolatile characteristic of the capacitive ionic memristor, which is determined by confined ion transport and sorption rates in the pores of polarized nanoporous carbons. There is no pronounced change in the memory time using a polyvinyl alcohol gel electrolyte demonstrating that ion diffusion in the bulk electrolyte plays only a minor role for the memory time (Figure S46).

4-terminal devices facilitate the understanding of the mechanism of resistance change induced by electrically-driven ion adsorption. In addition, 2- and 3-terminal devices as common types of ionic memristors were investigated in Figure 4h–i and Figure S47–48. In a 3-terminal device (Figure 4h), two large electrodes were used as gate (G) and source (S) electrodes (the voltage applied to G and S was denoted U_G) to effectively control ion adsorption. S electrode and a third small electrode (drain (D) electrode) form a working capacitor (W-Cap). After ten subsequent 1 V-pulses to G and S electrodes, the resistance of W-Cap showed a similar increasing trend as D-Cap in a 4-terminal device (Figure 4i and S47). For a C₁₂AmOMCl electrolyte, we observed a 47 % increase in the resistance of W-Cap in this 3-terminal device, which is lower than that for a 4-terminal device (85 %). However, the increase for the ChCl electrolyte (29 %) was higher than that for a 4-terminal device (15 %). The difference is due to the different resistance detected in 3- and 4-terminal devices (Figure 4b and h). The resistance changes in 2-terminal devices were also detected and compared for ChCl and C₁₂AmOMCl electrolytes (Figure S48), which further confirmed that the resistance changes resulted from electrically-driven ion adsorption. Moreover, we compared the memory time of 2-, 3-, and 4-terminal memristors (Figure S49 and Section 2 in the Supporting Information). These results demonstrate that ion confinement in nanoporous carbons provides a generalized strategic concept and platform for various memristive device architectures.

Electrically-Driven Concentration Change and Resistance Switching Behaviour

Different interactions of C₁₂AmOMCl and ChCl lead to the different switching behavior of the resistance and chloride anions remain mobile in both electrolytes. As it is challenging to monitor the local concentration of chloride anions spectroscopically, we selected salicylate anions as a model to investigate the concentration fluctuation induced by electric polarization, based on the characteristic ultraviolet (UV) absorption of salicylate anions (see the calibration curves in Figure S50). The physisorption of NaSal was compared with ChCl and C₁₂AmOMCl via in situ-Raman spectra (Figure 5a). We observed the physisorption behavior in ROX carbon, which is intermediate between the ChCl and C₁₂AmOMCl, due to the aromatic structure (hydrophobic interaction) of salicylate anions (electrostatic repulsion). The increase in resistance of D-Cap in a NaSal electrolyte (36 %) was also higher than that in NaCl (11 %) and ChCl (6 %) electrolytes, and lower compared to C₁₂AmOMCl electrolyte after applying ten 1 V-pulses to the M-Cap (Figure 5b).

The electrically-driven concentration fluctuation was evaluated in a cuvette cell (Figure 5c). Two ROX carbon electrodes and a separator were assembled inside the cuvette. The concentration of the bulk electrolyte was monitored by ultraviolet-visible (UV/Vis) spectroscopy after 3 days of equilibration time (Figure 5d). When the cell voltage was increased to 1 V (working electrode (WE) was positively polarized, and counter electrode (CE) was negatively polarized), salicylate anions in the bulk electrolyte were adsorbed by WE and there was a decreasing trend in intensity (concentrations of salicylate anions). When the cell voltage was decreased to −1 V, WE was negatively polarized, salicylate anions were desorbed and reinjected into the bulk solutions (i.e., an increase in concentrations, Figure 5d). The in situ-UV/Vis results indicate that the controlled ion adsorption can be controlled by electric signals. The electrically-driven concentration changes lead to resistance changes in the proposed capacitive ionic memristor devices, in response to the dynamic ion adsorption processes.